354 L. Bell — Absolute Wave-length of Light. 



The temperature was given by a thermometer in contact 

 with S a „ and 1^ of the micrometer equalled 0""28. In a com- 

 parison of two standards with such unequal coefficients of ex- 

 pansion as glass and speculum metal, the evil effects of radia- 

 tion should be at their maximum, but the preceding series, 

 including as it does all the experimental errors and showing 

 an extreme variation of but O 1 "^, leaves, I think, little to be 

 desired. 



The comparator was placed in a vault some six feet below the 

 level of the street, which was provided with thick double walls 

 with an air space between. This observing room enabled the 

 temperature to be kept down to a daily variation of less than 

 half a degree, the extreme range for several days being fre- 

 quently less than that amount. Before this vault in the new 

 Physical Laboratory was completed the comparator had been 

 placed in an upper room of one of the old buildings, where it 

 was well nigh impossible to keep anything like a constant 

 temperature, particularly since the heat was unavoidably par- 

 tially shut off during the night. Owing to this state of affairs 

 the measurement of the gratings on which my preliminary 

 wave-length was based, was made under difficulties and in 

 most of the series necessarily under a rising temperature. 

 Now when a glass standard is measured against a metal one, 

 glass being a notoriously bad conductor, and having a very 

 small coefficient of expansion, if any rise of temperature takes 

 place the length found for the glass will be too small, for re- 

 sponding less readily to a change it will be actually measured 

 at a low r er temperature. 



It therefore became necessary to re-measure the glass gratings 

 Nos. I and II, to eliminate this source of error, which was 

 done before the results for III and IY were obtained. These 

 gratings are very nearly 3 cm long and they were therefore com- 

 pared with successive triple centimeters of S a 2 until the fifteen 

 centimeter mark was reached. Grating I was first taken in 

 hand and six complete series of observations were obtained, 

 each micrometer reading being the mean of several, and the 

 extreme limits of temperature variation during the two days 

 occupied by the comparisons being 0°*3 C. The following 

 gives a summary of the results. 



5G = 15 cm S% -fi^-oy 



oG = 15 cm S a „ +21-5 | 

 5G=15-S\"+ 18-1 I Atl9 o. 9C 

 5G = 15 cm S a „ + 23-9 I Al iy y V " 

 5G = 15 cm S% + 22-6 | 

 5G = 15 cm S\ + 18-3 J 



